S. Hu, H. Qi, Z. Wang et al.
Environmental Science and Ecotechnology 30 (2026) 100682
2.1. Definitions
2.1.1. Definition of functional zones within PPPs We divided each plant into five functional zones to capture differences in production processes and carbon emissions. This zoning scheme adheres to Chinese national standards (i.e., the Code for the Design of Pulp and Papermaking Plants [37]) and was refined based on spatial patterns observed in high-resolution remote-sensing imagery. Areas not belonging to these five func- tional zones were classified as other built-up areas. The following list summarizes the functional roles, emission characteristics, and key visual features of PPP functional zones (see the Supplementary Fig. S1 for representative examples). Primary fiber stacking areas. These areas are used for the outdoor storage of raw primary fiber materials and serve as the upstream functional zone of the pulping process. They are asso- ciated with primary fiber pulping, which is characterized by high energy consumption and relatively high carbon emissions. These areas are typically rectangular or circular and are painted in yellow or dark brown. Recovered fiber stacking areas. These areas are used to store bundles of recovered fiber and appear as rectangular patterns. Recovered fiber pulping requires less energy and results in lower emissions compared with primary fiber pulping. Identification was based on spatial arrangement, spectral reflectance, and material color (typically yellow). Wastewater treatment areas . These areas contain sedimen- tation tanks, oxidation basins, and treatment pools. Water bodies vary in color depending on the treatment stage, with untreated wastewater appearing darker and treated effluent lighter. Emis- sions from this zone are mainly indirect. Structures are typically circular or square and form regular patterns. Thermal power plant areas . These areas provide on-site energy supply and constitute the primary source of direct carbon emissions due to fuel combustion. They are characterized by tall chimneys and cooling towers with linear, circular, or elliptical structures, and they exhibit distinct spatial features in remote-sensing images. Other stacking areas . These areas are used for outdoor storage of various materials and serve as a supporting role in the pro- duction system. Their emissions are relatively minor and mainly originate from material handling and short-distance trans- portation. In remote-sensing imagery, they appear as regular geometric patterns, most commonly rectangular, with diverse surface colors that distinguish them from surrounding areas. Other built-up areas . These areas include workshops, ware- houses, offices, and related structures. Emissions from this zone are predominantly indirect and mainly result from electricity consumption. They are characterized by densely clustered build- ings that form large, contiguous color blocks readily distinguish- able from those of other functional zones. 2.1.2. Definition of PPPs We defined PPPs based on the delineation of functional zones combined with plant product information as follows: Primary fiber pulp plants (PFPPs). Plants characterized by the presence of primary fiber stacking areas on the plant sites (Supplementary Fig. S2a). Recovered fiber pulp plants (RFPPs). Plants characterized by the presence of recovered fiber stacking areas and the absence of primary fiber stacking areas (Supplementary Fig. S2b). Heavyweight paper product manufacturing plants (HPPMPs). Plants that lack both primary fiber and recovered fiber stacking areas, with production mainly focused on heavyweight paper products, excluding tissue and specialty papers (Supplementary Fig. S2c).
Carbon accounting. First, we delineated the boundaries of each PPP and applied contour extraction techniques to identify the re- gions of interest (ROIs) corresponding to plant areas from remote- sensing imagery. We then used a semantic segmentation model based on DeepLabv3 + to segment and identify functional zones within each plant, enabling quantification of the area associated with each functional region. Simultaneously, we cleaned the textual data to remove noise and analyzed it using a BERT model to generate plant-classification results. We subsequently integrated the outputs of the image-based and text-based analyses at the decision level. Specifically, we used image-based information mainly to distinguish between primary fiber pulp plants and recovered fiber pulp plants, whereas we used textual data to classify other types of PPPs. Based on the final plant categories, we developed an area-based carbon accounting model for each PPP, using functional-zone measure- ments and corresponding carbon-emission data. CER potential assessment. Based on the PPP area, geographical location, and solar radiation data, we developed a mechanistic model to evaluate the CER potential of rooftop PV power genera- tion. We then employed the Kucherenko index to conduct a global sensitivity analysis (GSA) and quantify the influence of the model input variables and their interactions on the outputs. The Kucherenko index uses a quasi-random Sobol sequence with high spatial sampling efficiency to ensure that the control variables are mutually independent. This not only allows the first-order effects of individual factors to be assessed but also captures interactions between variables. When the sensitivity index exceeds 0.1, the factor is considered highly sensitive [36]. This analysis enabled us to identify the parameters with the greatest impact on CER po- tential. On this basis, we systematically assessed the CER potential of PPP rooftop PV systems across a range of parameters. Fig. 1. Framework of this study. The left panel shows the carbon accounting process. Image- and text-based classification results are integrated to determine functional zones and categorize pulping and papermaking plants (PPPs) into five types, enabling the estimation of plant-level carbon emissions for 720 PPPs. The right-hand panel illustrates the assessment of the potential to reduce carbon emissions, where the generation of photovoltaic (PV) power is estimated, and global sensitivity and sce- nario analyses are conducted to evaluate the potential to reduce carbon emissions at the plant level under different panel lengths. BERT: bidirectional encoder represen- tations from the transformers.
3
Made with FlippingBook interactive PDF creator